Every agent run gets its own box, isolated by default

Firecracker microVMs for the code your agents write. Start one in a single call, pause and resume it, on one simple monthly plan.

Start building for free
Compute

Compute for agents, run by us

Sandboxes, workspaces, build machines and your own lab hardware behind one API. No servers to set up.

Sandscale compute, US East and EU. Sandboxes, workspaces and build hosts on our machines in N. Virginia and Frankfurt, ready with one API call.

Your lab, connected. Plug in your own Linux or macOS machines with real devices and claim them per run.

Sandboxes

Firecracker microVMs for agent code. Run commands, move files, stream logs, snapshot, pause and resume.

Workspaces

A persistent cloud computer for your coding agent, with your repo, terminals and SSH. Keep working from any device.

Build hosts

Lease a clean build machine for a set time, sync your tree and run over SSH with a short-lived certificate. Wiped on release.

Lab hosts

Connect your own Linux or macOS machines with security keys, TPMs, debug probes and boards, and claim them per run.

Why Sandscale

Why teams and developers run agents on Sandscale

Agents need a place to run code nobody has read

Running it on Sandscale

sbx_9a1c finished in its own box

See inside
every agent run

Claude Code runs under your policy through its hooks: each tool call is allowed, blocked or sent to a person. Codex and Gemini CLI report every step over OpenTelemetry. Pause or stop any session from the dashboard.

See sandboxes →
Gemini CLIsbx_9a17
Summarise last week’s failed builds
sbx_9a17 added to context
read 31 build logs
recorded over OpenTelemetry
Claude Codesbx_9a14
Fix the failing payment tests
sbx_9a14 added to context
ran npm test · 42 passed
blocked: curl … | sh
waiting for lee@ to approve a PR
Codexacme-api
Lint the web app and fix what you can
acme-api added to context
ran npm run lint
edited 6 files
recorded over OpenTelemetry
Claude Codesbx_9a18
Add retries to the upload client
sbx_9a18 added to context
paused by lee@

Policy
and audit

Decide which templates, sizes, run times and agents each project may use, and which calls need a person first. Every sandbox, workspace, lease and change lands in a hash-chained audit log you can export.

How we secure it →
Policy › payments
Templates
python-3.12, node-22
Largest sandbox
4 vCPU
Longest run
2 hours
Network
Off
Per person
5 at once
Agents
Claude Code, Codex
Ask first
git push, gh pr create
Block
piped installs, rm -rf /
Audit logHash-chained · export
TimeWhoActionTargetEntry
14:02:11maya via Claude Codesandbox.createsbx_9a149a1c3f
14:02:14maya via Claude Codevault.injectSTRIPE_TEST_KEY4be210
14:02:31maya via Claude Codetool.blockedcurl … | shd07a91
14:02:58sam via Codexworkspace.connectacme-api71c0e4
14:03:20lee@acme.devapproval.grantedgithub.create_pull_requeste2f5b8
14:03:41lee@acme.devsandbox.pausesbx_9a180c93ad
14:04:02schedulerbuildhost.releaselease_7f2c · wiped5a6e17
Pricing

Start free. Scale when your agents do.

(Every plan runs on the same isolated microVMs.)

One plan, every kind of box

Sandboxes, workspaces and build hosts all draw from your plan's monthly hours. Pro gives one developer room to work; Team adds the controls a company needs; Enterprise adds your own cloud and your own keys.

Your planSandscale
Sandboxes
Workspaces
Build hosts
Lab hosts
Vault
Same isolation and a basic audit log on every plan. Paid plans add size, network access and scale.
Free
for trying it out
$0/month
 
100 vCPU-hours2 at onceNetwork off
Start free
Pro
for one developer
$20/mo billed annually
or $25/mo billed monthly
750 vCPU-hoursWorkspacesBuild hostsVault
Start with Pro
Team
for teams
$40/member/mo
billed annually, or $48 monthly
1,000 vCPU-hours / memberSSO and rolesAudit export
Start with Team
Enterprise
for larger organisations
Custom
Billed annually
SCIM and BYOKDedicated capacityYour own cloud
Talk to us
Compare all features
FreeProTeamEnterprise
Compute
Members112 or moreCustom
Included vCPU-hours a month1007501,000 per member, pooledCustom
Running at once220100Custom
Largest boxUp to 2 vCPUUp to 8 vCPUUp to 16 vCPUUp to 64 vCPU on request
Usage beyond your hoursNoTurn on or offTurn on or offPer agreement
RegionsUS and EuropeUS and EuropeUS and EuropeUS and Europe
Products
Sandboxes✓✓✓✓
Pause and resume✓✓✓✓
SnapshotsNo✓✓✓
WorkspacesNo✓✓✓
Build hostsNoLeases up to 4 hLeases up to 12 h✓
Lab hostsNoNo✓✓
Agent CloudNo✓✓✓
Object storage (US now, Europe coming soon)No50 GB500 GBCustom
Network and security
NetworkOffAllowlistAllowlistAllowlist
Vault for secretsNo✓✓✓
Audit logBasicBasicWith exportWith export
Approvals and policies, projectsNoNo✓✓
DLP on prompts and tool calls (Enterprise)NoNoNo✓
Passkeys and multi-factor sign-in✓✓✓✓
Team and enterprise
Organisations, invitations, roles, groupsNoNo✓✓
SSO (SAML and OpenID Connect)NoNo✓✓
SCIM, verified domains, custom rolesNoNoNo✓
Your own encryption keysNoNoNo✓
Custom sign-in domain and brandingNoNoNo✓
Dedicated capacityNoNoNo✓
Bring your own cloud (AWS; Azure and Google Cloud coming soon)NoNoNo✓
Support
SupportDocs and communityEmailEmailSupport terms in your agreement
Need more than your plan?

On paid plans you choose what happens when your included hours run out: turn usage beyond them on, or keep it off and your sandboxes and workspaces wait for next month. On Enterprise you can also run on your own cloud.

Need bigger machines? Enterprise can run boxes of up to 64 vCPU on request.

Your cloudSandscale
AWS
AzureSoon
Google CloudSoon
Bring your own cloud on Enterprise.
Product

Everything a run needs

Start it, watch it, stop it. From your agent, your terminal or the dashboard.

Every run, listed and audited

Start, pause, snapshot and stop sandboxes from the API, the sandscale CLI or MCP. Every action lands in a tamper-evident audit log.

One platform for agent compute

Sandboxes, workspaces, build hosts and lab hardware run on the same platform, managed from one dashboard and billed as one monthly plan.

Secure by default

Its own kernel per run, the network off until your allowlist opens it, secrets kept in Vault and never in prompts, and a kill switch for your organisation.

Built for coding agents

Connect over MCP from Claude Code, Codex or your own agent framework. Each agent gets its own box, and you see what each one is doing.

Give every agent
a box of its own

Start free with 100 vCPU-hours a month. Upgrade when your agents need more.

A Firecracker microVM for every runNo network by defaultPause and resume with state keptVault keeps secrets out of promptsSSO, roles and audit logBring your own cloudWorks over MCP and the CLI